Why a waterfall (not a single source).
No single data vendor covers everyone. Apollo is great for US SaaS, weak in APAC. LinkedIn Sales Navigator has every contact but no email. ZoomInfo nails enterprise. A waterfall queries them in sequence and stops at the first hit — best coverage, lowest cost per row.
The 6 columns of a good enriched row.
- Company — name, domain, headcount, industry, location, funding round.
- Contact — full name, title, seniority, department.
- Email — verified deliverable (NeverBounce or DropContact pass).
- LinkedIn — direct URL, not company page.
- Tech stack — BuiltWith / Wappalyzer pull.
- AI summary — 2 sentences: what they do + what just happened.
The Clay build.
One table. Columns 1–4 are sequential waterfalls (Apollo → ZoomInfo → LeadMagic). Email validation is a single column that runs only when an email exists. The AI summary is a Claude column with the prompt: "In two sentences, describe what {{company}} sells and what's different about them based on their homepage." Skip rows with low confidence — better to drop than to bluff.
Cost per row, end-to-end.
$0.08–$0.15 per fully enriched row at 1,000 rows/month — cheaper than a single SDR seat by orders of magnitude, and the data refresh runs nightly.
The Clay table export and waterfall config ship in the weekly. Pair this with WF-01 Signal-based Outbound for the full loop.